how house price dynamics and credit constraints affect the ......home equity comprises about half of...
TRANSCRIPT
How House Price Dynamics and Credit Constraints affect the
Equity Extraction of Senior Homeowners
Stephanie Moulton, John Glenn College of Public Affairs, The Ohio State University
Donald Haurin, Department of Economics, The Ohio State University
Samuel Dodini and Maximillian Schmeiser, Federal Reserve Board
Disclaimer:
The research reported herein is being performed pursuant to a grant from the
MacArthur Foundation as part of the “How Housing Matters” Research Competition
and with funding from The U.S. Department of Housing & Urban Development’s
Office of Policy Development and Research (PD&R). The opinions and
conclusions expressed are entirely those of the authors and do not represent the
opinions of the MacArthur Foundation or HUD.
The views expressed in this paper are those of the authors and do not necessarily
represent the views of the Federal Reserve Board, the Federal Reserve System, or
their staffs.
1. Empirical Modeling of Reverse Mortgage Borrower Behavior • Take-up of HECMs (and other equity extraction products among seniors)
• HECM technical default (property tax and insurance default)
• HECM loan terms, withdrawal behaviors and termination outcomes
• Equity extraction (including HECMs) and longer term credit outcomes
2. Survey of Counseled Seniors • Longer term well-being of HECM borrowers
• May 2014-July 2015, about 2,000 respondents: (1) current HECM
borrowers, (2) terminated HECM borrowers, and (3) seniors who sought
counseling but did not get a reverse mortgage.
3. Post Origination Monitoring Pilot • RCT design; financial planning and reminders after closing
• Launched January, 2015
Research Program (2012-2017)
Motivation
Home equity is an important part of a senior household’s financial portfolio:
Approximately 80% of households over the age of 62 own their homes
(Poterba et al. 2011)
Home equity comprises about half of seniors’ median net wealth (2013
SCF)
Home equity is a significant source of retirement funds for baby boomers
(Lusardi and Mitchell 2007; Wolff 2007)
Different options to extract equity:
Selling and moving
Cash-out refinancing, second liens or HELOCs
Reverse mortgages- federally insured HECMs
Research Questions
What factors are associated with seniors’ extraction of equity through
various channels, including a reverse mortgage?
• Do neighborhood house price dynamics and credit conditions
differentially affect originations by channel? • Do homeowners in credit constrained areas respond differentially to an
increase in house prices than homeowners in non-constrained areas?
• Do high minority share neighborhoods respond differently than low
minority share neighborhoods? (50+% minority vs 90+% white)
• Is the share of equity extracted through particular channels
differentially associated with foreclosure rates among extractors?
Previous studies have generally focused on the broader population and exclude
reverse mortgages (Hurst and Stafford 2004; Mian and Sufi 2011; Do 2012;
Bhutta and Keys 2014; Duca and Kumar 2014; LaCour-Little et al. 2014). Further,
they do not jointly model different channels of equity extraction.
Equity Extraction
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
0.000
0.010
0.020
0.030
0.040
0.050
0.060
0.070
2004 2005 2006 2007 2008 2009 2010 2011 2012
Mean Equity Extraction Origination Rate using any Channel as a Proportion of the
Population 62 and older, by Year
Equity Loan
Equity Extraction by Channel
0.000
0.020
0.040
0.060
0.080
0.100
0.120
0.140
0.000
0.010
0.020
0.030
0.040
0.050
2004 2005 2006 2007 2008 2009 2010 2011 2012 HE
CM
Sh
are
of
Ori
gin
ati
on
s
Mean Equity Extraction Origination Rate as a Proportion of the
Population 62 and older, by Year
HELOC Cashout Refinance
Closed-End Second HECM
HECM Share
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Geographic Variation (U.S.): HELOCs
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean HELOC Origination Rate as a Proportion
of the Population 62 and older, 2004-2012
Geographic Variation (U.S.): Cash-Out Refinancing
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean Cash-Out Refinancing Origination Rate as a Proportion
of the Population 62 and older, 2004-2012
Geographic Variation (U.S.): HECMs
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean HECM Origination Rate as a Proportion
of the Population 62 and older, 2004-2012
Geographic Variation
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean HELOC Origination
Rate as a Proportion
of the Population 62 and
older, 2004-2012
Geographic Variation
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean Cash-Out Refinancing
Origination Rate as a
Proportion of the Population
62 and older, 2004-2012
Geographic Variation
Source: Author’s calculations from HUD HECM data and the New York Fed’s Consumer Credit Panel Data
Mean HECM Origination
Rate as a Proportion of the
Population 62 and older,
2004-2012
Theoretical Expectations
• House prices & channel of extraction Higher house prices, wealth effect (+ all channels); also relax borrowing constraint,
allowing access to products with lower LTV requirements (+ HELOC and HECMs)
As house prices are increasing, preserve option to extract again in future periods;
not lock into high initial cost product (- HECM)
As house prices are decreasing, lock in house values today (+HECM)
• Credit conditions & channel of extraction Supply side credit availability (+ forward originations)
Household credit history, credit standards can create binding constraint (- HELOC)
Household liquidity constraints (+ HELOC and HECM)
• House price*credit constraints Credit constrained borrowers may be more responsive to house price increases;
originate through channels with relaxed credit constraints (+ cash out refinancing)
• Neighborhood demographics (minority share) & channel of extraction
Endowment effects, different levels of explanatory factors in minority areas
Differential responses to explanatory factors due to financial literacy, experience
Data Sources & Sample
Data Sources 1. New York Federal Reserve’s Consumer Credit/Equifax Panel (CCP) database,
2004-2012
• 4th quarter, 62 or older, +12 million credit profiles
• Aggregated to ZIP code and year
2. HUD HECM database and actuarial database, 2004-2012
• 697,772 originations
• Aggregated to ZIP code and year
3. CoreLogic, ZIP code level data, 2004-2012
• House price and HPI for non-distressed sales
4. IRS (SPEC) Tax data, 2004-2012
• Elderly tax filing data by ZIP code, median adjusted gross income (AGI)
5. ACS data, ZIP code level demographic indicators, 2005-2010
• Data from the 2000 U.S. Census to interpolate values for 2003 and 2004
Sample Limit to ZIP codes within CBSAs with HPI data across all years, and to those with at
least 30 CCP records for consumers aged 62 or older in a given year =
• 5,495 ZIP codes (covers about 45% of the full population)
Resulting sample = 39,596 unique ZIP code and year combinations
Empirical Model: Seemingly Unrelated Regression
Yzt = β0 + β1HPzt + β2CCzt + β3Xzt + α1IChannel,zt + γm + δt + uzt
Y= (1) HELOC origination rate
(2) Cash-out refinancing origination rate
(3) Second lien origination rate
(4) HECM origination rate
For each ZIP code z at time t
HP = house price dynamics (median repeat sales price, HPI growth rate)
CC = credit conditions (credit approval rate, credit utilization rate, credit score, etc.)
X = control variables (median AGI, mortgage debt, median age, black, Hispanic, etc.)
I = interest rate for extraction channel (averaged over the year within the ZIP code)
γ = CBSA fixed effects
δ = year fixed effects
• Alternative specifications include interactions, HP*CC
• Estimate subsample regressions in ZIP codes with high levels of racial homogeneity
Allow error terms of 4
equations to be correlated,
common component and
random component
Findings: Overall
SUR Estimates, % Population 62 + Equity Extraction Method, 2004-2012 (Select Variables Shown)
Values = regression coefficient divided by the mean percentage of originations
HELOC Cash-Out Refinance
Closed-End Second
HECM
Variable b/ȳ b/ȳ b/ȳ b/ȳ Median Real Repeat Sales Price (ln) 0.006 *** 0.001 *** -0.001 *** 0.001 ***
HPI Growth Rate, Positive 0.598 *** 0.408 *** -0.145 -2.289 *** HPI Growth Rate, Negative -0.018 -0.295 ** 0.168 1.016 *** Credit Approval Ratio (All) 0.734 *** 1.096 *** 0.598 *** 0.144 **
Median Credit Score 0.002 *** -0.006 *** -0.002 *** -0.005 ** Median Revolving Credit Utilization Rate 1.115 *** -0.145 0.144 0.621 **
Median IRS AGI (Monthly, thousands) 0.070 *** -0.076 *** -0.059 *** -0.176 *** Black (Share of Population) -0.328 *** 0.708 *** 0.040 1.305 ***
Hispanic (Share of Population) -0.171 *** 0.410 *** -0.043 0.075 Year & CBSA Fixed Effects Y Y Y Y
R-Squared 0.542 0.239 0.215 0.467 Dependent Variable Mean 0.024 0.008 0.007 0.002
*** p<0.01, ** p<0.05, * p<0.1
Findings: House Price Growth*Credit Constraints
Credit Constraint Interactions with Positive and Negative HPI Growth Rate
Panel A: Credit Score Interactions
% Δ HELOC % Δ
Cash-Out
% Δ
Second % Δ HECM % Δ Rate
At mean credit score (784)
0.01 Increase in HPI Rate 0.9257 0.2285 -0.0763 -2.1108 0.493
0.01 Decrease in HPI Rate -0.1375 -0.2538 0.2379 0.6427 -0.065
One standard deviation (20 points) below the mean credit score (763)
0.01 Increase in HPI Rate 0.2821 0.5881 -0.2253 -2.3539 0.150
0.01 Decrease in HPI Rate 0.0819 -0.3088 0.0110 1.6224 0.063
One standard deviation (20 points) above the mean credit score (803)
0.01 Increase in HPI Rate 1.5331 -0.1110 0.0644 -1.8814 0.836
0.01 Decrease in HPI Rate -0.3446 -0.2020 0.4520 -0.2821 -0.192
Findings: Geographic Subsample Regressions
0.04
0.0176
0.0106 0.0083
0.0035
0.0415
0.0262
0.0067 0.007
0.0016
00.0050.01
0.0150.02
0.0250.03
0.0350.04
0.045
AnyOrigination
HELOC Cash-outRefinance
Closed-EndSecond
HECM
Origination Rate as a Proportion of the Population 62 and older, 2004-2012
All ZIPs >50% Black >90% White
HELOC Cash-Out Second HECM
Minority Area Difference - + + +
Endowment Effect + - - -
Behavioral Response + - + +
Findings: Geographic Subsample Regressions
SUR Estimates, % Population 62 + Equity Extraction Method, 2004-2012, by Geographic Subsamples
HELOC HELOC Cash-Out Refinance
Cash-Out Refinance
High
Minority Low
Minority High
Minority Low
Minority Variable b/ȳ b/ȳ b/ȳ b/ȳ
Median Real Repeat Sales Price (ln) 0.004 *** 0.007 *** 0.000 0.000 HPI Growth Rate, Positive 0.020 0.637 *** 0.447 0.041
HPI Growth Rate, Negative -0.111 -0.542 *** -0.826 ** 0.091 Credit Approval Ratio (All) 0.790 ** 0.618 *** 1.226 *** 0.630 ***
Median Credit Score 0.002 * 0.001 -0.001 -0.010 *** Median Revolving Credit Utilization Rate 0.457 2.000 *** 0.594 -1.881 ***
Median IRS AGI (Monthly) 0.120 ** 0.072 *** -0.120 ** -0.030 *
Black (Share of Population) -0.565 ** 0.342 0.561 ** -0.087 Hispanic (Share of Population) -0.841 ** 0.086 0.159 0.282
Year & CBSA Fixed Effects Y Y Y Y R-Squared 0.485 0.524 0.408 0.179
Dependent Variable Mean 0.018 0.026 0.011 0.007 *** p<0.01, ** p<0.05, * p<0.1
Findings: Geographic Subsample Regressions
SUR Estimates, % Population 62 + Equity Extraction Method, 2004-2012, by Geographic Subsamples
Closed-End
Second Closed-End
Second HECM HECM
High
Minority Low
Minority High
Minority Low
Minority Variable b/ȳ b/ȳ b/ȳ b/ȳ
Median Real Repeat Sales Price (ln) 0.000 -0.001 * 0.002 *** 0.000 *** HPI Growth Rate, Positive -0.478 0.089 -3.914 *** -2.169 ***
HPI Growth Rate, Negative -0.260 0.447 ** 0.070 2.594 *** Credit Approval Ratio (All) 0.739 * 0.446 *** 0.343 -0.208 **
Median Credit Score -0.001 -0.006 *** 0.000 0.003 * Median Revolving Credit Utilization Rate -0.282 0.861 0.250 -0.062
Median IRS AGI (Monthly) -0.035 -0.047 *** -0.326 *** -0.189 *** Black (Share of Population)
Hispanic (Share of Population) Year & CBSA Fixed Effects Y Y Y Y
R-Squared 0.296 0.208 0.617 0.517 Dependent Variable Mean 0.008 0.007 0.004 0.002
*** p<0.01, ** p<0.05, * p<0.1
Findings: Foreclosure Rates by Extraction Channel
0.021 0.027
0.033 0.033
0.023 0.017
0.022
0.000
0.020
0.040
0.060
0.080
0.100
0.120
2004 2005 2006 2007 2008 2009 2010
Fore
clo
sure
Rat
e
Origination Year
Foreclosure Rates as of 2013(Q4), by Origination Cohort and Extraction Channel
Any Extraction
HELOC
Cash-Out Refinance
Closed-End Second
HECM
Findings: Foreclosure Rates by Extraction Channel
OLS, Proportion of Extractors Foreclosing as of
Q42013, By Origination Cohort (Select Years) 2004 2006 2007 2010
Cash-Out Refinancing 0.041*** 0.045*** 0.063*** 0.034***
Closed-End Second 0.002 0.005 0.012 0.001
HECM 0.004 0.014 -0.016 -0.004
% w/Mortgage Past Due 0.079* 0.080** 0.095*** 0.126***
Credit Utilization Rate 0.010** 0.004 0.0197** -0.025***
Credit Score (100s) -0.016*** -0.022*** -0.013*** -0.030***
Credit Approval Ratio -0.018 -0.056*** -0.045*** -0.051***
Constant 0.148** 0.290*** 0.148** 0.447***
Observations 4,555 4,646 4,586 2,828
CBSA Fixed Effects Yes Yes Yes Yes
R-squared 0.203 0.234 0.231 0.243
*** p<0.01, ** p<0.05, * p<0.1
For 2007 Originations:
A 10 percentage point
increase in cash-out
refinancing is associated
with a 19% increase in the
foreclosure rate among
extractors (0.0063/0.033)
A 10 point increase in
median credit score is
associated with an 4%
decrease in the
foreclosure rate
(0.0013/0.033)
If HECMs were replaced
by cash-out refinancing,
the foreclosure rate could
have been 12% higher
Conclusions
• Significant differences in the determinants of the origination of a home equity
extraction loan by channel:
Variation in responsiveness to house prices by channel
Variation in responsiveness to credit conditions by channel
Variation in responsiveness to house prices in credit constrained areas by
channel
• Differences in channel use in high vs. low minority areas is due in part to
differences in endowments and differences in behavioral responses.
Minority areas less responsive to house price increases and decreases (to
originate HELOCs or HECMs)
Low-minority areas less likely to use cash-out refinancing or HECMs when
credit conditions improve
• The usage of particular channels in an area is significantly associated with
foreclosure rates among extractors.
An increase in the share of extractions through cash-out refinancing is
associated with significantly higher foreclosure rates
While HECMs are originated in similarly credit constrained areas, HECM
origination share in an area is not significantly associated with foreclosure
Discussion & Implications
• As of April 2015 HECM lenders must assess a borrower’s “ability to pay” and
follow minimum credit, debit and affordability standards. • In a prior paper, we estimate a 6 percent reduction in HECM volume due to the
credit portion of the policy, based on the proportion of households who would
“fail” the criteria and be unable to afford an escrow for taxes and insurance.
• We estimate that the policy could reduce tax and insurance default by as much as
40 percent.
• Using the characteristics of the households who we estimate would be excluded
from HECMs based on the policy, we predict the probability that they would instead
obtain another equity extraction loan. We estimate that these excluded
households would be very unlikely to have originated a HELOC, but would be
more likely to have originated a second lien or cash-out refinance loan instead
of a HECM.
• To the extent that HECM loans have built in protections (e.g., insured against
negative equity), these households may turn to more “risky” alternatives.
Moulton, S., D. Haurin and W. Shi. 2015. An Analysis of Default Risk in the Home Equity
Conversion Mortgage (HECM) Program. Journal of Urban Economics
Thank You!
1. Empirical Modeling • HECM terminations & default
• Take-up of HECMs
• HECM loan terms and withdrawal behaviors
2. Survey of Counseled Seniors • Longer term well-being of HECM borrowers
• May 2014-May 2015, about 2,000 respondents: (1) current HECM
borrowers, (2) terminated HECM borrowers, and (3) seniors who sought
counseling but did not get a reverse mortgage.
3. Post Origination Monitoring Pilot • RCT design; financial planning and reminders after closing
• Launched January, 2015
Research Program (2012-2016)
Table 1: Descriptive Statistics for Model Variables, Full Sample (N=39,596)
mean sd min max
HELOC Origination Rate 0.0244 0.0209 0.0008 0.2670
Cash-out Refinance Origination Rate 0.0078 0.0079 0 0.1470
Closed-End Second Origination Rate 0.0066 0.0073 0 0.0771
HECM Origination Rate 0.0019 0.0020 0 0.0293
Median Repeat Sales Price (ln) 12.4800 0.5600 10.2800 14.9700
HPI Growth Rate, Positive 0.0460 0.0762 0 0.7840
HPI Growth Rate, Negative 0.0367 0.0568 0 0.5690
HELOC ZIP-level Interest Rate 0.0579 0.0126 0.0200 0.1225
First Mortgage ZIP-level Interest Rate 0.0538 0.0088 0.0250 0.0825
Closed End Second ZIP-level Interest Rate 0.0668 0.0102 0.0206 0.1161
Average HECM MSA-level Interest Rate 0.0561 0.0004 0.0425 0.0657
Credit approval rate (All) 0.6720 0.0836 0.2310 1.0000
Median Credit Score 783.58 20.18 634 820
Median Revolving Credit Utilization Rate 0.0793 0.0478 0.0152 0.5760
Past Due Mortgage Rate 0.0165 0.0198 0 0.2310
Bankruptcy Rate 0.0090 0.0090 0 0.1360
Foreclosure Rate 0.0027 0.0047 0 0.0760
Revolving Debt to Income Ratio (1 yr lag) 0.0204 0.0136 0 0.5670
Share of Population with Mortgage (1 yr lag) 0.3370 0.1090 0.0502 1.0000
Median Mortgage Debt to Median Sales Price (1 yr lag) 0.3720 0.1560 0 2.4420
Median Monthly Mortgage Payment (1 yr lag) 0.8840 0.3380 0.1360 3.5630
Median IRS AGI (Monthly) 3.5520 1.3520 0.4170 8.3330
Median Age of Seniors with Credit Files 72.4600 2.3100 65 84
Black (share of population) 0.0980 0.1460 0 0.9810
Hispanic (share of population) 0.1300 0.1510 0 0.9750
Our other papers: • Haurin, D., C. Ma, S. Moulton, W. Shi, M. Schmeiser, and J. Seligman. (Forthcoming). Spatial
Variation in Reverse Mortgages Usage: House Price Dynamics and Consumer Selection.
Journal of Real Estate Finance and Economics.
• Moulton, S., D. Haurin and W. Shi. 2015. An Analysis of Default Risk in the Home Equity
Conversion Mortgage (HECM) Program. Journal of Urban Economics (forthcoming)
• Working Papers: (1) Reverse mortgage choice and the influence of counseling; (2) Dynamic
model of reverse mortgage outcomes; (3) Seniors’ accuracy of home valuation
Research Program (2012-2016)
Reverse Mortgage 101
• In the U.S, the federally insured Home Equity Conversion Mortgage (HECM)
comprises 95% of the market. Small, but potentially growing market.
• Extract equity from the home through a mortgage that does not become due
until the last borrower sells the home, moves out permanently, or dies, as long
as the borrower meets the obligations of the mortgage note
• Obligations include living in the home as primary residence, pays
property taxes, homeowners insurance, homeowners association dues
and assessments, and maintains the home.
• No payments on the loan are required during the life of the loan. Money
borrowed, plus associated interest and fees, are added to the balance due
that continues to grow over time (mortgage “in reverse”)
• Line of Credit
• Tenure or Term (similar to annuity)
• Lump Sum Distribution
• Some combination of the above
Reverse Mortgage Debt
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
1 5 10 15 20
$ A
mo
un
t
Time (Years)
Lump Sum
Expected
Home Value
Maximum Claim Amount (home value at closing)= $225,000
Initial Principal Limit = $125,000
Reverse Mortgage Debt
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
1 5 10 15 20
$ A
mo
un
t
Time (Years)
Available
Credit Line
Expected
Home Value
Credit Line or
Term/Tenure
Payments
Maximum Claim Amount (home value at closing)= $225,000
Initial Principal Limit = $125,000
Source: CFPB 2012
Source: CFPB 2012
http://www.ginniemae.gov/doing_business_with_ginniemae/issuer_resources/ginni
enet/NewIssuerTrainingPresentations/20131122_HMBS_Panel.pdf